Determination of displacements on built-up steel beams during experiments by image processing


Ozsariyildiz S. S. , DÜĞENCİ O. , Sulev E.

JOURNAL OF CONSTRUCTIONAL STEEL RESEARCH, cilt.179, 2021 (SCI İndekslerine Giren Dergi) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 179
  • Basım Tarihi: 2021
  • Doi Numarası: 10.1016/j.jcsr.2021.106560
  • Dergi Adı: JOURNAL OF CONSTRUCTIONAL STEEL RESEARCH

Özet

In this study, an image processing technique is developed for measuring the displacements during flexural loading experiments of structural beams as an alternative to measurements by traditional gauging devices. In this context, four-point bending tests were carried out on three built-up (welded) steel beams, one of them without stiffening plates and the others with stiffening plates. In the experiments, while the load-deflection values in each loading step were taken using load cells and LVDTs, a video recording of the experiments was simultaneously taken. The load-deflection curves were comparatively constructed using both the values measured with LDVTs and the data obtained by image processing. In order to match the data obtained using the point flow method of image processing technique to the gauged measurements, the scale coefficients in pixel / mm unit were calculated. The mean absolute error (MAE) and root mean square error RMSE error values were computed to determine the efficacy of the scale coefficients. The MAE and RMSE values for the beam without the stiffening plates turned out to be 3.42 and 4.52, while they were 1.29 and 1.82 for the beams with the stiffening plates. According to these results, the higher the lateral torsional buckling, the more difficult it is to determine the verification coefficient. The accuracy of the image processing method developed herein for determination of in-plane displacements can be further improved easily for experiments involving out-of-plane displacements due to local buckling also if the recordings of more cameras from different angles are included in the analyses.